Why the First Three Seconds Decide the Whole Video
On TikTok, Reels and Shorts, the algorithm doesn't measure whether someone liked your video — it measures whether they watched it, and for how long. That single signal, completion rate, is the strongest input into whether a short-form platform shows your video to more people or lets it die at a few hundred views. Viewers decide whether to keep watching in the first one to three seconds, before your setup, your logo, or your slow "hey guys, welcome back" intro ever gets a chance. If the hook doesn't land immediately, the rest of the script — no matter how good — never gets watched.
This is the single biggest difference between a script written for a video platform and a script written for anything else. A blog post can ease into its point. A video script cannot. The hook has to either promise a specific, believable payoff ("the one email subject line mistake killing your open rate") or create a specific, curious gap ("I found out why my open rate was stuck at 12% and it wasn't the subject line at all") — both work because they're concrete, not because they're clever. Vague hooks like "let's talk about email marketing today" give a viewer nothing to stay for, and they scroll.
The Four-Beat Structure: Hook, Setup, Payoff, CTA
Every short-form script that holds attention follows some version of the same four beats, just with different amounts of time in each one depending on total length:
Hook (first 0–3 seconds). One line, spoken and mirrored on-screen as text, that states or implies the payoff. This is the only beat that should never be cut for time — if your video is running long, the hook is the last thing to trim.
Setup (next few seconds, skipped entirely on very short videos). One sentence of context: why this matters, who it's for, or what situation makes the payoff relevant. On a 15-second video there usually isn't room for this beat at all — you go straight from hook to payoff.
Payoff (the middle, and the longest beat). This is where you deliver on what the hook promised. On longer videos this can be broken into two or three numbered points ("first... second... third..."), which also gives editors natural cut points for on-screen text and b-roll. On a 15-second video, stick to exactly one point delivered cleanly — trying to cram three ideas into 9 seconds means none of them land.
CTA (final 3–5 seconds). A single, specific ask — follow, comment a keyword, save, or share — not three asks stacked on top of each other. Platforms increasingly reward comments and saves as engagement signals distinct from likes, so a comment-bait CTA ("comment 'MORE' and I'll send you the full list") often outperforms a generic "follow for more" close.
Pacing: How Many Words Actually Fit
A commonly used rule of thumb for natural, conversational spoken pace is around 2.5 words per second, or roughly 150 words per minute — slower than an auctioneer, faster than a formal presentation. That means a 15-second video has room for roughly 35–40 spoken words total, a 30-second video for about 70–80 words, and a 60-second video for around 140–150 words. Once you subtract the hook and CTA, that leaves surprisingly little room in the middle — which is exactly why most beginner scripts run long: they try to fit a blog post's worth of ideas into a video's worth of seconds.
The fix isn't to talk faster. It's to cut ideas, not words per idea. If your script is running over, remove a whole point from the payoff rather than trying to compress every sentence. A 30-second video that delivers one idea clearly outperforms one that rushes through three.
On-Screen Text Isn't Optional
A large share of short-form video is watched with the sound off — on a phone in a waiting room, on a train, in bed next to someone asleep. If your hook and key points only exist as spoken audio, you lose every muted viewer in the first three seconds, which on some platforms is a majority of the audience. Every beat of your script should have a matching on-screen text line: not a transcript of every word, but the compressed, punchy version — usually five words or fewer — that captures the beat's core idea. Think of on-screen text as the "headline" for each beat, and the spoken line as the elaboration underneath it.
Adapting One Script Across TikTok, Reels and Shorts
The four-beat structure works on all three platforms, but the details worth adjusting differ. TikTok's discovery algorithm leans especially hard on completion rate and rewatches, which is why very short (15–21 second), tightly looped videos often outperform longer ones with the same content — a video that gets watched twice in 18 seconds sends a stronger signal than one watched once in 45. Reels leans more on trending audio and visual editing than pure voice-over, so writing on-screen text that works with the sound off (and the audio muted or swapped for trending music) matters even more there. YouTube Shorts rewards a strong verbal hook in the first two to three seconds, and because Shorts sits inside a platform built around subscriptions, a mid-video re-hook ("but here's the part most people miss") followed by an explicit subscribe CTA at the end tends to perform better than it would on TikTok or Reels.
In practice, this means: write the script once with the generator above, then adjust the CTA and, if needed, the on-screen text density before posting to each platform — rather than writing three completely separate scripts from scratch.
Common Mistakes That Kill Watch Time
The most common script mistake is a slow, generic opening — "hey guys, in this video I'm going to talk about..." — which burns the entire hook window without giving the viewer a reason to stay. The second most common mistake is stacking too many CTAs at the end ("like, comment, follow, and check the link in bio") which dilutes all of them; pick one action per video. The third is ignoring the word budget and trying to explain something genuinely complex in 15 seconds — if a topic needs more than 60 seconds to do justice, that's a signal to either split it into a short series or move it to a longer-form video instead of cramming it.
Related tools: Caption & Hook Generator · Hashtag Generator · Content Repurposing Planner